Background
Congenital Hyperinsulinism in Infancy (CHI) is a rare disorder causing severe debilitating hypoglycaemia, usually presenting in infancy [
1,
2]. Hypoglycaemia due to CHI can have a deleterious impact on early life brain function, with several cohorts reporting adverse neurodevelopmental outcomes in a third to a half of patients [
3‐
6]. The frequency of hypoglycaemia-related brain injury in the CHI population as a whole has not reduced despite optimisation of diagnosis and treatment over the last decade. The burden of morbidity in CHI continues to be a major problem for individuals and health care professionals; therefore, a greater focus is required on understanding variations in disease severity.
Genetic understanding of CHI has progressed rapidly with a significant proportion of CHI found to have underlying genetic causes, most frequently mutations in the K-ATP channel genes,
ABCC8 and
KCNJ11 [
7,
8]. K-ATP channel genotyping has stratified treatment protocols of focal and diffuse CHI with paternal heterozygosity most commonly associating with focal CHI and maternal heterozygous, homozygous or compound heterozygous mutations in
ABCC8/
KCNJ11 associating with diffuse disease [
2]. Although paternal heterozygosity has a higher predilection for focal CHI, additional investigation such as 18-fluoro-dopa PET-CT scanning is necessary to localise the lesion in focal CHI; a significant proportion, as many as half with paternal heterozygous mutations in some reports may have diffuse CHI [
9] which could be explained by dominant inheritance or inability to identify a maternal mutation in recessively-inherited disease.
It is recognised that pancreatectomy, either lesionectomy for focal lesions or subtotal pancreatectomy for severe diffuse CHI is a well-established treatment choice for CHI. However, increasingly there is a shift to conservative medical management particularly in the case of diffuse CHI which is traditionally treated by near total pancreatectomy. Indeed, some children with focal CHI in the head of the pancreas proximal or abutting the bile duct may benefit from conservative treatment due to the nature of the surgical complexity involved. In our centre, the frequency of patients (with K-ATP and non-K-ATP channel gene mutations) undergoing pancreatic surgery as a proportion of new patients referred to the service has reduced from 18% in 2007–2008 to 6–7% in 2014–2015.
A number of case reports of spontaneous resolution of disease have been reported [
10‐
12], mostly in those without known genetic mutations, while cohort studies in different countries have characterised surgical outcomes only [
4,
7,
8,
13]. Long-term conservative treatment with diazoxide and octreotide without requirement for pancreatic surgery has also been reported in patients with and without K-ATP channel gene mutations [
12,
14,
15]; however these observations do not offer insight into the evolution of disease severity and if treatment response improves or worsens over time. Therefore, disease trajectories of medically treated K-ATP CHI remain poorly understood. It is important to understand the trends in severity of CHI to modify and individualise the intensity of medical therapy. Here we have studied a cohort of patients with K-ATP CHI treated by medical therapy to examine outcomes of disease in follow up assessments.
Methods
The aims of our study were to assess variation in intensity of treatment in children with K-ATP CHI over time, and to review outcomes of medically treated K-ATP CHI patients in follow up assessments.
A cohort of patients with K-ATP CHI (mutations in ABCC8/KCNJ11) treated by medical therapy (n = 21) was identified from a group of patients (n = 404) in a specialist centre for CHI between April 2006 and July 2016, with local Research Ethics approval. Genetic investigations were performed in 269 patients only within the cohort. In the remainder, genetic investigations were not performed because CHI resolved in early infancy or patients remained on low dose diazoxide. In those undergoing genetic testing, 71 patients had mutations in ABCC8/KCNJ11, 10 patients had mutations in other genes related to CHI (HNF4A, GCK, HADH, GLUD1) and 10 patients had variants of uncertain clinical significance. Within the group of 71 patients with ABCC8/KCNJ11 mutations, 39 patients underwent pancreatic surgical treatment (subtotal pancreatectomy or focal lesionectomy); patients who were not surgically treated, i.e. medically treated (n = 21) were recruited to the study. Eleven patients who were also medically treated were not recruited because they either presented between January 2016 and July 2016, or insufficient clinical information was available in follow up.
The diagnosis of CHI was made in patients presenting to this centre using well-established criteria [
1,
2]. Patients underwent rapid K-ATP channel gene mutation analysis as per protocol, as previously reported [
10]. Variants either previously reported or considered likely to be pathogenic were included in the cohort. One variant reported as pathogenic in our patient but classified elsewhere as being a variant of uncertain significance was also included.
The diagnosis of focal CHI was made on the basis of a paternal heterozygous mutation in
ABCC8/KCNJ11 and confirmed by identification of a solitary lesion in the pancreas during 18-fluoro-dopa PET-CT scanning [
2]. Those with no clear foci were diagnosed as diffuse CHI. Diffuse CHI was also presumed if the patient had maternal heterozygous, homozygous, or compound heterozygous mutations in
ABCC8/KCNJ11, for which 18-fluoro-dopa PET-CT scans were not performed. Patients with
ABCC8/KCNJ11 mutations who required either lesionectomy for focal CHI or subtotal pancreatectomy for diffuse CHI were excluded from the cohort. Patients who underwent pancreatic biopsy or minimal resection while continuing medical therapy were included in the cohort.
Treatment variations were made on clinical grounds and individualised to patient need. Oral diazoxide was used as first line treatment, while somatostatin agonists (SSRA, octreotide, somatuline) were used as second line treatment. Carbohydrate supplements to increase energy content of milk and polyunsaturated fatty acids (PUFA) used in the management of diazoxide responsive CHI were considered as food supplements and did not preclude inclusion to the cohort [
16]. The dose of Eicosapentaenoic acid (EPA) component of omega-3 fatty acid was allowed in a range of 240–480 mg per day. Responsiveness to diazoxide as treatment for CHI was determined by noting satisfactory glucose profiling and fasting tolerance as described previously [
16]. Responsiveness to SSRA was also determined in a similar manner.
Children had resolution of CHI if treatment was minimised and withdrawn completely with maintenance of satisfactory glucose profiles (95% values >3.5 mmol/L) on home glucose monitoring or subcutaneous continuous glucose monitoring (CGM) [
10,
16]. To achieve resolution of CHI, satisfactory fasting tolerance was mandatory with end of fast blood glucose >3.0 mmol/L, suppressed insulin concentrations and blood ketones >1.0 mmol/L measured by point of care testing and/or laboratory analysis of 3 hydroxybutyrate. Follow-up consisted of telephone reviews every 2 weeks for the first 4 months, followed by clinic reviews at 4 monthly intervals by a multi-disciplinary team including a clinician, two specialist nurse practitioners, two dietitians, a speech and language therapist and a clinical psychologist. At each review, glucose profile was assessed and medication adjusted accordingly. Children who demonstrated resolution of CHI were reviewed in clinic appointments every 6 months by a clinician and specialist nurse practitioner without wider multi-disciplinary team input. Annual home blood glucose profiles were assessed to determine glycaemic status and to ensure continuing euglycaemia. Oral glucose tolerance testing was not performed routinely in all children undergoing spontaneous resolution, in the absence of information regarding long-term utility and difficulty in administering the test in young children. Instead, home blood glucose profiling was assessed and correlated with symptoms of hypoglycaemia and hyperglycaemia. Pancreatic biopsy was not routinely undertaken in patients enrolled in the cohort. However, for patients in whom a pancreatic biopsy was undertaken as a partial resection, the tissue was analysed for characteristics of focal and diffuse CHI [
17].
In addition to glycaemic outcomes in follow up assessment, the Vineland Adaptive Behavior Scales, version II (VABS-II), a questionnaire completed by parents was used to assess adaptive functioning in the domains of communication, daily living skills, social skills and motor skills after age 1.5 years (
http://www.pearsonclinical.com/). Information was also obtained on the prevalence of seizures and delayed development in clinical assessment [
3]. Auxology parameters were reviewed at the 2 year follow up assessment and measurements were converted to Standard Deviation Scores (SDS) [
18]. Statistical analysis was performed by IBM-SPSS version 23.0 (IBM incorporated, New York, USA); Mann–Whitney test was performed to test differences between non-parametric independent variables while paired t-tests were used to test difference between paired samples.
Discussion
Our study of young patients with K-ATP CHI suggests that resolution of CHI occurs in a significant proportion (71%) of those safely managed by conservative medical treatment. Resolution may not occur in all patients in prolonged follow up, but there is reduction in the intensity of treatment for hypoglycaemia, suggesting a trend of reducing severity of disease over time.
Our findings of reducing severity in both recessively or dominantly inherited
ABCC8/KCNJ11 mutations extend the recognised theme that dominant mutations may be mild [
19] and that resolution can occur in a few children with recessively inherited disease [
11,
20]. This notion is also commensurate with observations in large cohorts where patients with homozygous and compound heterozygous mutations may be medically managed without need for pancreatic surgery [
7]. While it is recognised that the natural history of CHI may become clinically more manageable, our report provides objective and systematic evidence for this prevailing notion. Our findings also provide much needed prognostic information about the disease trajectory of K-ATP CHI and guidance for clinicians to re-evaluate severity at successive intervals and reduce medication as necessary.
We accept that patient numbers are relatively small and that only five patients with compound heterozygous and homozygous mutations represented severe diffuse medically treated CHI. However, patient numbers are not small for a rare disease drawn from a relatively large group of patients with genetic and non-genetic CHI over a 10 year period. Nonetheless, replication in other international cohorts would be helpful to prove the strength of association. Larger cohorts and international databases would be required to determine factors associated with reduction in severity as the number of patients in our cohort were too few (n = 7) to hypothesise mechanisms of disease resolution in CHI caused by recessively inherited mutations.
Only six children in this cohort remained on long-term medication. Two of these patients had missense mutations affecting
KCNJ11 residue p.R206. Three other patients tested in Exeter had mutations at this residue and had congenital hyperinsulinism that persisted for between 21 months and >3 years. The
ABCC8 p.R526C mutation was reported in a patient who required treatment up to the age of 6 years [
21]. However, a genotype:phenotype correlation is not absolute since the
ABCC8 p.I1512T mutation was found in another patient tested in Exeter whose hyperinsulinism remitted within a few days of birth.
In our study, we have provided genetic information on the type of K-ATP channel gene mutations in CHI patients. However, we have not investigated genotype predictions of natural history phenotype as in-silico predictions are unreliable in establishing pathogenicity and have not been tested in model predictions of disease trajectory. As experience in medical management of patients with K-ATP CHI accumulates worldwide, our study suggests the need to generate phenome databases to derive genotype-assisted prediction models of disease prognosis.
Although patients in our cohort had reducing severity, the neurodevelopmental phenotype was no different to previous cohorts [
3,
5,
6]. This is likely to reflect adverse impact of hypoglycaemia in early life [
3] and not likely to reflect the impact of continuing hypoglycaemia, as home glucose monitoring had been satisfactory in all patients. Further strength comes from the observation that the majority of the most severe patients, i.e. those with homozygous and compound heterozygous mutations had normal neurodevelopmental outcomes.
We did not observe deterioration in oral feeding with treatment reduction and disease resolution. The majority of children in this cohort were orally fed; those requiring gastrostomy tube feeding improved oral feeding over time. Therefore, treatment withdrawal or reduction was not associated with the collateral effect of increasing reliance on gastrostomy tube feeding.
Although we have reported a reduction in disease severity in the natural history and progression of genetic forms of CHI, we have been unable to find markers at presentation that could predict the resolution of disease. Therefore, it follows that CHI should be treated aggressively at the outset as recommended [
1,
22], but with regular monitoring in follow-up to reduce treatment dosage, where feasible. The reduction in treatment intensity is not only a responsive management strategy, but also potentially reduces the significant harm to patients from excessive doses and prolonged exposure to medications with recognised toxic adverse effect profiles. We would recommend telephone and/or electronic communication every 2 weeks for the first 4 months to understand trends in home glucose profiles and drug response, followed by 4 monthly clinic reviews to assess the need for dose reduction. We would also suggest annual review of therapy for those remaining on treatment for longer than a year. Although we did not find patients experiencing relapse of hypoglycaemia in the relatively short duration of follow up, we would suggest on-going monitoring for the risk of hypoglycaemia, particularly during illness episodes for at least 2 years.
One criticism to adopt a step down treatment approach is the exposure to the potential risk of hypoglycaemia. However, the frequency of adverse neurodevelopment in our cohort was no different in those between resolution and persistence of CHI and no different than previous cohorts [
5,
6]. The frequency of adverse neurodevelopment in the medically treated group has not been compared directly with the frequency in patients treated surgically in our cohort, although comparison of our data with other cohorts suggests a similar prevalence [
4]. If early onset hypoglycaemia is the most important determinant of later life adverse neurodevelopment [
3], it is unlikely that the small risk of hypoglycaemia from a proposed reduction in therapeutic intensity would be more detrimental. Nonetheless, it would be advisable to weigh up risks and benefits when offering treatment de-escalation choices to parents of children with CHI.
In our study of natural history outcomes, we did not evaluate glucose tolerance as part of the assessment of glycaemic outcomes, unlike other studies following pancreatectomy [
23]. However, the utility of glucose tolerance testing at a young age in patients with resolving CHI not requiring surgery has not been established. Nonetheless, it would be important to evaluate formal glucose tolerance in older children and adolescents with resolved CHI to investigate the probability of evolving hyperglycaemia and diabetes.
Acknowledgements
The authors are grateful to research nurses and clinical colleagues at Central Manchester University Hospitals NHS Trust and the Manchester Biomedical Research Centre.